CSCE Operating Systems Deadlock. Qiang Zeng, Ph.D. Fall 2018

Transcription

1 CSCE Operating Systems Deadlock Qiang Zeng, Ph.D. Fall 2018

2 Previous Class What is Deadlock? How to detect it? Dealing with deadlock: Prevention Avoidance Detection CSCE 311 Operating Systems 2

3 Outline Barrier Problem Readers-Writers Problem Review of Synchronization Issues of lock-based programming CSCE 311 Operating Systems 3

4 Deadlock A set of processes is deadlocked when each process in the set is blocked awaiting an event that can only be triggered by another blocked process in the set Some Slides Courtesy of Dr. William Stallings CSCE 311 Operating Systems 4

5 Potential Deadlock I need quad C and D I need quad B and C I need quad D and A I need quad A and B CSCE 311 Operating Systems 5

6 Actual Deadlock HALT until D is free HALT until C is free HALT until A is free HALT until B is free CSCE 311 Operating Systems 6

7 Resource Categories Reusable can be safely used by only one process at a time and is not depleted by that use processors, I/O channels, main and secondary memory, devices, and data structures such as files, databases, and semaphores Consumable one that can be created (produced) and destroyed (consumed) interrupts, signals, messages, and information in I/O buffers CSCE 311 Operating Systems 7

8 Example of Deadlock: Memory Request Space is available for allocation of 200Kbytes, and the following sequence of events occur: P1... Request 80 Kbytes;... Request 60 Kbytes; P2... Request 70 Kbytes;... Request 80 Kbytes; Deadlock occurs if both processes progress to their second request CSCE 311 Operating Systems 8

9 Example of Deadlock: waiting for messages Consider a pair of processes, in which each process attempts to receive a message from the other process and then send a message to the other process: S1 = 1; s2 = 1; P1: P(s1) P(s2) P2: P(s2) P(s1) CSCE 311 Operating Systems 9

10 Resource Allocation Graph There is a circle in the graph, which indicates deadlock CSCE 311 Operating Systems 10

11 Resource Allocation Graph describing the traffic jam CSCE 311 Operating Systems 11

12 Conditions for Deadlock Mutual Exclusion Hold-and-Wait No Pre-emption Circular Wait A process cannot access a resource that has been allocated to another process a process may hold allocated resources while awaiting assignment of others no resource can be forcibly removed from a process holding it a closed chain of processes exists, such that each process holds at least one resource needed by the next process in the chain CSCE 311 Operating Systems 12

13 Dealing with Deadlock Three general approaches exist for dealing with deadlock: Prevent Deadlock adopt a policy that eliminates one of the conditions Avoid Deadlock make the appropriate dynamic choices based on the current state of resource allocation Detect Deadlock attempt to detect the presence of deadlock and take action to recover CSCE 311 Operating Systems 13

14 Deadlock Condition Prevention Mutual Exclusion Hold and Wait Avoiding mutual exclusion is not realistic Countermeasure: require that a process request all required resources at once; blocking the process until all requests can be granted simultaneously CSCE 311 Operating Systems 14

15 Deadlock Condition Prevention No Preemption Countermeasure: if a process holding certain resources is denied a further request, that process must release its original resources and request them again Circular Wait Countermeasure: define a linear ordering of resource numbers; if a process has been allocated a resource of number R, then it may subsequently request only those resources of numbers following R in the ordering. Why does this work? Think about the Resource Allocation Graph CSCE 311 Operating Systems 15

16 Deadlock Avoidance Deadlock prevention breaks one of the deadlock conditions through rules, which are defined before execution, while deadlock avoidance is enforced during execution A decision is made dynamically whether the current resource allocation request will lead to an unsafe state Requires knowledge of future process requests We will examine some examples CSCE 311 Operating Systems 16

17 Example State of a system consisting of 4 processes and 3 resources Allocations have been made as follows CSCE 311 Operating Systems 17

18 Determination of a Safe State P2 requests one of R1 and one unit of R3 Should this request be granted? Banker s algorithm: assume this request is granted, then check whether the resulted state is safe A state is safe if there is at least one sequence of resource allocations that satisfies all the processes needs Is this a safe state? CSCE 311 Operating Systems 18

19 P2 Runs to Completion Old Available vector (0, 1, 1) + Resources released by P2 (6, 1, 2) = Updated available vector(6, 2, 3) CSCE 311 Operating Systems 19

20 P1 Runs to Completion Old Available vector (6, 2, 3) + Resources Released by P1 (1, 0, 0) = Updated available vector(7, 2, 3) CSCE 311 Operating Systems 20

21 P3 Runs to Completion Thus, the state defined originally is safe CSCE 311 Operating Systems 21

22 Determination of an Unsafe State P1 requests for one more R1 and one more R3 The request should not be granted, because it leads to an unsafe state CSCE 311 Operating Systems 22

23 Deadlock detection CSCE 311 Operating Systems 23

24 Recovery strategies Kill one deadlocked process at a time and release its resources Kill all deadlocked processes Steal one resource at a time Roll back all or one of the processes to a checkpoint that occurred before they requested any resources, then continue Difficult to prevent indefinite postponement CSCE 311 Operating Systems 24

25 Recovery by killing processes CSCE 311 Operating Systems 25

26 CSCE 311 Operating Systems 26

27 Dining Philosophers: failed solution with deadlock # define N 5 void philosopher (int i) { while (TRUE) { think(); take_fork(i); take_fork((i+1)%n); eat(); /* yummy */ put_fork(i); put_fork((i+1)%n); } } CSCE 311 Operating Systems 27

28 Dining Philosophers: failed solution with deadlock # define N 5 void philosopher (int i) { while (TRUE) { think(); take_fork(i); take_fork((i+1)%n); eat(); /* yummy */ put_fork(i); put_fork((i+1)%n); } } CSCE 311 Operating Systems 28

29 Dining Philosophers: failed solution with deadlock # define N 5 void philosopher (int i) { while (TRUE) { think(); take_fork(i); take_fork((i+1)%n); eat(); /* yummy */ put_fork(i); put_fork((i+1)%n); } } CSCE 311 Operating Systems 29

30 Dining Philosophers solution with numbered resources Instead, number resources First request lower numbered fork # define N 5 void philosopher (int i) { while (TRUE) { think(); take_fork(lower(i)); take_fork(higher(i)); eat(); /* yummy */ put_fork(lower(i)); put_fork(higher(i)); } } 2 CSCE 311 Operating Systems

31 Dining Philosophers solution with numbered resources Instead, number resources... Then request higher numbered fork # define N 5 void philosopher (int i) { while (TRUE) { think(); take_fork(lower(i)); take_fork(higher(i)); eat(); /* yummy */ put_fork(lower(i)); put_fork(higher(i)); } } 2 CSCE 311 Operating Systems

32 Dining Philosophers solution with numbered resources Instead, number resources... Then request higher numbered fork # define N 5 void philosopher (int i) { while (TRUE) { think(); take_fork(lower(i)); take_fork(higher(i)); eat(); /* yummy */ put_fork(lower(i)); put_fork(higher(i)); } } 2 CSCE 311 Operating Systems

33 Dining Philosophers solution with numbered resources Instead, number resources... One philosopher can eat! # define N 5 void philosopher (int i) { while (TRUE) { think(); take_fork(lower(i)); take_fork(higher(i)); eat(); /* yummy */ put_fork(lower(i)); put_fork(higher(i)); } } 2 CSCE 311 Operating Systems

34 Summary What is Deadlock? How to detect it? Dealing with deadlock: Prevention Avoidance Detection CSCE 311 Operating Systems 34